Neural circuits are the processing units of the brain, formed by networks of thousands or millions of nerve cells that transform information. The retina is the only neural circuit in the central nervous system (CNS) that can be removed intact and its entire natural input delivered to hundreds of thousands of cells with millisecond precision. It is also the only neural circuit of the CNS that can potentially be replaced by an electronic prosthesis in cases of disease. A substantial barrier to this therapeutic goal is a detailed understanding of the internal processing of information in the retina - the neural code. To understand the two sequential layers of processing in the retinal circuit, we will take a divide-and- conquer approach to understanding the neural code by simultaneously recording at two different levels in the circuit using a novel combination of techniques. Visual scenes are projected from a video monitor onto the photoreceptor layer of an isolated, intact salamander or mouse retina. An infrared two-photon laser-scanning microscope is used in conjunction with a voltage-sensitive dye to record the activity of ~ 100 interneurons. To do this, we will use second harmonic generation imaging, a technique that selectively images the membrane potential of neurons using dyes with minimal pharmacological effects. Simultaneously, an array of sixty electrodes is used to record action potentials in the output cells of the retina that comprise the optic nerve. We will use measured visual responses at two layers of the retina to separately model outer and inner retinal processing. We will also use simultaneous recordings from interneuron and ganglion cell populations to test whether the inner retina compresses the visual scene by reducing redundancy contained in its synaptic input. These results will have immediate applicability to the emerging field of retinal prostheses. The objective of a retinal prosthesis system is to treat prevalent diseases such as age-related macular degeneration and retinitis pigmentosa by replacing the function of the damaged retina with a high resolution electronic circuit. Measurements of the retinal neural code will be directly useful for incorporation into subretinal and epiretinal prostheses systems.